{"id":"W2954903132","doi":"10.1109/icse.2019.00107","title":"CRADLE: Cross-Backend Validation to Detect and Localize Bugs in Deep Learning Libraries","year":2019,"lang":"en","type":"article","venue":"","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":191,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; MNIST database; Implementation; Deep learning; Artificial intelligence; Software bug; Software; Reliability (semiconductor); Machine learning; Anomaly detection; Key (lock); Software engineering; Programming language; Operating system","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004364083,0.0001472787,0.0001798797,0.0001823292,0.0001137505,0.000583025,0.0004836755,0.00008189766,0.0001583262],"category_scores_gemma":[0.0003237622,0.0001413169,0.00002504278,0.0005610946,0.00004084995,0.001351148,0.0006186223,0.0003081748,0.0001862108],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003882718,"about_ca_system_score_gemma":0.00003350112,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001048419,"about_ca_topic_score_gemma":0.00002404841,"domain_scores_codex":[0.9986227,0.0001342458,0.0002293904,0.0004817449,0.0002323062,0.0002995707],"domain_scores_gemma":[0.9992263,0.0002659225,0.00006605847,0.0003146901,0.0000373342,0.00008965228],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00003277383,0.000009279475,0.1405851,0.0000346841,0.000008741144,0.00001411349,0.002149054,0.7144665,0.0004166909,0.03483779,0.00002041471,0.1074248],"study_design_scores_gemma":[0.0009115057,0.0001942947,0.06501482,0.00004248334,0.000003063408,0.00001693638,0.0001500552,0.9182135,0.002292144,0.007038987,0.005674099,0.0004480799],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2919794,0.00004874278,0.7044094,0.0002978017,0.0001907298,0.0001503267,8.398507e-8,0.0001700424,0.002753536],"genre_scores_gemma":[0.8730202,0.000004132909,0.1253803,0.0002275343,0.00003606219,0.000006298475,0.000001518158,0.00001391245,0.001310042],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.5810408,"threshold_uncertainty_score":0.5762736,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.007898350116497276,"score_gpt":0.2634367416265574,"score_spread":0.2555383915100602,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}